Since about 1999, I have maintained my own mail servers, and probably have too many email accounts to mention (my own domains, of which there are several, gmail, work, which is O365, old work addresses, other companies email servers I maintain, etc), and I note that over the past week or so, spam that has not triggered filters has skyrocketed on all of my accounts (which are not related, e.g. usersnames and domains are different).

Am wondering, typically I see ups and downs in spam from one service or another (e.g. "gee, this week gmail sure is not doing a good job" or "wow, my own filters are doing great, not a spam message in three weeks" etc), but this past seven days have seen an across the board exponential increase in pure spam that filters just are not filtering.

This leads me to also wonder about the state of the art in machine learning spam filtering. There was a time when there seemed to be great interest or promise in it, but quick searches only seem to revel the same (older) posts about training simple models, which seem to marginally preform better than bayes (aka spamassassin/amavisd).

Just an endnote, not interested in arguing about running my own mail servers, or complaining, truly just curious if other mail admins have seen this huge uptick, or are thinking about ML for spam detection.